Building and Enuironment, Vol. 25, No. 4, pp. 309-316, 1990. 036()-IJ2J/90 noo+o.

oo
Printed in Great Britain. © 1990 Pergumon Press pie.

Literature Review on Thermal Comfort in

Transient Conditions

J. L. M. HENSEN*

The co11ve11tio11a/ theory of thermal comfort i11 co11di1io11s dwrac1eris1ic for dweili11gs and offices
(for example. that of Fa119er) a.mmres steady-siate conditions. Yet tliermal co11di1io11s i11 b11ildi11gs
are seldom steady, due 10 the i111erac1ia11 betweC/1 building stmcwre, clima1e, occ11pa11cy, and
HVAC system. This article reviews work 011 thermal comfort specifically 1111dertaken to examine
what uarialio11s i11 indoor 1empera111res may be acceptable.
Fol/owi11g an accou/11 ofman's 1/iermoreg1ifa1ory system, some experimemal.fi11di11gs 011 periodic
am/ 011 ramp (or drift) varia1io11 i11 room temperature are presented.
!1 is concluded that rhe results for cyclic varia1io11s uphold the presellt ASH RAE standard. but
those for drifts may not.

1. INTRODUCTION which a large percentage of occupants (generally at least

80%) with given personal parameters will regard the
THERMAL COMFORT is generally defined as that con-
environment as acceptable.t Most work related to ther-
dition of mind which expresses satisfaction with the ther-
mal comfort has concentrated on steady-state conditions.
mal environment (1]. Dissatisfaction may be caused by
This is expressed by the fact that only one of the above
the body being too warm or cold as a whole, or by un-
standards [7] also specifies limits for changing environ-
wanted heating or cooling of a particular part of the
mental parameters (for T only).
body (local discomfort). 0

Because of the thermal interaction between building

From earlier research (2--4] we know that thermal com-
fort is strongly related to the thermal balance of the body.
This balance is influenced by:
e Environmental parameters like: air temperature (T.)
and mean radiant temperature (T,),t relative air vel-
ocity (v) and relative humidity (rh)
e Personal parameters like : activity level or metabolic
rate (M) (units: 1met=58 W m- 2) and clothing
thermal resistance (/c 1) (units: 1 clo = 0.155 m 2 K
w-1).
Extensive investigations and experiments involving
numerous subjects have resulted in methods for pre-
dicting the degree of thermal discomfort of people
exposed to a still thermal environment. The most well
known and widely accepted methods are: (1) Fanger's
"Comfort Equation" and his practical concepts of "Pre-
dicted Mean Vote" and "Predicted Percentage of Dis-
satisfied" (2]; (2) the J. B. Pierce two-node model of
human thermoregulation (4, 5]. With these methods sev-
eral thermal comfort standards [l, 6-8] have been estab-
lished during the past decade. These standards specify
environmental parameter ranges (i.e. comfort zones) in
• Eindhoven University or Technology, Group FAGO HG
11.77, P.O. Box 513, 5600 MB Eindhoven, The Netherlands."
t T, is often combined with T, to fonn operalive temperature
(T0 =aT. +(l-a)T,wherea <I).
t For exnmple, CSO [I) recommends for ligh'I, mainly sed·
entary activity during winter conditions (he.ating period) : "(a)
The operative temperarure sltal/ be beMeen 20 and 24°C (i.e.
22±2°C). (b) ... "; and during summer conditions (cooling
period) : "(a) The operative temperarure shall be between 23 011d
26°C (I.e. 24.5± I.SOC). (b) ..• ":
structure, occupancy, climate and HVAC system, pure
steady-state conditions are rarely encountered in
practice. For example, Madsen [9] found indoor tem-
perature fluctuations between 0.5 and 3.9°C (during 24
h with a constant set point) which depended on the com-
bination of heating and control system.
Sometimes it may even be advantageous to allow the
environmental conditions to change. This was dem-
onstrated in a field experiment [IO] where it was found
that decreasing the acceleration heating of the room ther-
mostat in a dwelling resulted in a lower fuel consumption.
This led however to considerably increased variations in
indoor temperature, but it was not clear at the time
whether these fluctuations would be acceptable or not to
inhabitants.
This is the background to the present literature study
on thermal comfort in transient conditions. We know
that temperjture is the most important environmental
parameter with respect to thermal comfort, so this study
focused mainly on the effects of changes in temperature
and mainly in homes, offices, etc.
In Section 2 man's thermoregulatory system is dis-
cussed so as to show the interaction between man,
building and HY AC system. Our present understanding
of human thermoregulatory mechanisms however is not
sufficient for us to predict with confidence our response
, to time-varying stimuli and recourse must be had to
controlled tests. The results of such work on cyclically
varying temperatures are present in Section 3.1 and on
other types of changes in the following section. Finally
in Section 4 some conclusions towards practical appli-
cations are made.
309
)
'I
310 J. L. M. Hensen
te11perat11re
sennotion
controllar
I.
. feedback skin
thenooreceptors
alamants
control lad body shell
system
uxternal
disturb•ncc
adjust-nt of
)IC?at cxcluln c:
control vasomotion
sweating
actions
clothing
-------+ signal path ------t> heat trensf er path
Fig. l. Schematic diagram of autonomic and behavioural temperature regulation in man (modified from
Hensel [11]) .
2. MAN'S THERMOREGULATORY SYSTEM
The human body produces heat (principally by metab-
olism- i.e. oxidalion of food elements), exchanges heat
with the environment (mainly by radiation and con-
vection) and loses heat by evaporation of body fluids.
During normal rest and exercise these processes result in
average vital organ temperatures near 37°C. The body's
temperature control system tries to maintain these tem-
peratures when thermal disturbances occur. According to
Hensel (11], who studied a vast amount of literature
on the subject, man's thermoregulatory system is more
complicated and incorporates more control principles
than any actual technical control system. It behaves
mathematically in a highly non-linear manner and con-
tains multiple sensors, multiple feedback loops and mul-
tiple outputs.
Figure I shows some basic feature's of man's thermo-
regulatory system. The controlled variable is an in-
tegrated value of internal temperatures (i.e. near the cen-
tral nervous system and other deep body temperaLUres)
and skin temperatures. The controlled system i.s influ-
enced by internal (e.g. internal heat generation by exer-
cise) and external (e.g. originating from environmental
heat or cold) thermal disturbances. External thermal djs-
turbances are rapidly detected by thermoreceptors in the
skin. This enables the lhermoregulatory system to act
before the disturbances reach the body core. Important
in this respect is the fact that the thermoreceptors in the
skin respond to the temperature as well as to the rate of
a temperature change. According to Madsen [12) the
latter is actually done by sensing heat flow variations
through the skin.
Autonomic thermoregulation is controlled by the
hypothalamus. There are different autonomic control
actions such as adjustment of : heat production (e.g. by
shivering) , internal thermal resistance (by vasomotion ;
i.e. control of skin blood flow), external thermal resist-
ance (e.g. by control of respiratory dry heat loss), water
secretion and evaporation (e.g. by sweating and res-
piratory evaporative heat loss). The associated tern-
thenial
COlllfort
hypothal811ns
internal
thenoorecepto'NI
body core autonomic
raguJ.ation
1nterna1
d11turb1nce
adjustraent of
heat production:
aetabol1sm
shiverinq
voluntary 1>0ve111ents
peratures for these autonomic control actions need not
necessarily be identical nor constant or dependent on
each other.
Besides autonomic thermoregulation there is also
behavioural thermoregulation with control actions such
as active movement and adjustment of clothing. Accord-
ing to Hensel [I I], behavioural thermoregulation is
associated with conscious temperature sensation as well
as with thermal comfort or discomfort. The difference
between temperature sensation and thermal comfort is
that temperature sensation is a rational experience that
can be described as being directed towards an objeclive
world in terms of "cold" and "warm". Thermal comfort
on the other hand is an emotional experience which
can be characterised in tenns of " pleasant" and "un-
pleasant". As Mcintyre [3] points out, the meaning of
words like "pleasant" and· comfortable' do not have an
absolute value, but will be relative to experience and
expectation.
Hensel (11] found that temperature sensations
(especially local cold sensations) depend mainly on the
activity of thermoreceptors in the skin whereas thermal
comfort or discomfort reflects a general state of the ther-
mo regulatory system (though this does not imply that
changes in thermal comfort are always slower than
changes in thermal sensation, as will be seen later on).
The condition of thermal comfort is therefore sometimes
defined as a state in which there are no driving impulses
to correct the environment by behaviour [13]. This is a
more objective definition than the ISO definition.
According to Mcintyre (3) it is conventional to treat
overall thermal discomfort (a subjective condili.on) in
term.s of thermal sensatjon (an objective quantity). This
may be justifiable in case of steady-state conditions how-
ever probably not when transient conditions have to be
judged. The difference between thermal comfort and tem-
perature sensation during changing environmental con-
ditions was clearly demonstrated by experiments of
Gagge et al. [14]. They exposed subjects for one hour to
neutral them1al conditions (29°C), then a step change to
a much colder (17.5°C) or warmer (48°C) environment
 Literature Review on Thermal Comfort
for a two hour exposure, which was followed by a step
back to neutral conditions. On entering the cold con-
ditions there were immediate reports of cold sensations
and discomfort. On returning to the neutral environment
discomfort almost immediately disappeared, while tem-
perature sensations lagged considerably behind the com-
fort reports and did not return to neutral for all subjects
during the one hour post-exposure period. The transient
exposures to the hot environment showed much the same
responses. On entering the hot conditions there were
immediate reports on warm sensations and discomfort.
On re-entering the neutral conditions discomfort dis-
appeared rapidly however more slowly than in the case
of the cold to neutral step. The temperature sensations
showed an overshoot with some initial reports of slightly
cool.
In the past much work has been done aimed at finding
practical methods for predicting the effects of a particular
thermal environment in terms of comfort or discomfort.
Reviews and summaries of this were made by Hardy [15],
Fanger [2], Benzinger (13], Mcintyre [3] and ASHRAE
[16]. From these references it is clear that there is much
evidence (from steady-state experiments) for cold dis-
comfort being strongly related to mean skin temperature
and that warmth discomfort is strongly related to skin
wettedness caused by sweat secretion. These relations are
the basis for methods like Fanger's [2] Comfort
Equation and the work of Gagge et al. [4, 5]. In a recent
evaluation by Doherty and Arens [17] it was shown that
these models are accurate for humans involved in near-
sedentary activity and steady-state conditions.
From the fact that the skin thermoreceptors not only
sense temperature but also the rate of temperature change
and that thermal comfort depends on an integrated value
of central and peripheral temperatures, it may be con-
cluded that skin temperature alone is unlikely to be an
adequate index for cold discomfort in transient
conditions. Because sweat secretion reflects the general
state of the thermoregulatory system, skin wettedness is
probably a more adequate predictive index for warmth
discomfort in transient conditions than mean skin tem-
perature is for cold discomfort. No experimental proof
of this has been found however. From these observations
one may conclude that the above mentioned essentially
steady-state methods are probably not adequate for pre-
dictions regarding thermal comfort in transient con-
ditions.
A number of models for simulation of the dynamic
behaviour of man's thermoregulatory system have been
developed in the past. A well known example is the model
of Stolwijk (18] which was later expanded by Gordon
(19]. In this model the human body is divided into a
large number of segments (originally 24 and in Gordon's
version 140) linked together via the appropriate blood
flows. Each segment represents volume, density, heat
capacitance, heat conductance, metabolism and blood
flow of a certain part of the body. The temperature and
rate of change of temperature of each segment is available
as an input into the controlling system, and any effector
•With triangular changes, peak-to-peak amplitude .1.T••••
cycle frequency CPH and rate of temperature OT/fit are related
according to :oT/Ot=2·CPH·.1.TP'•Kh-'. ".(
311
output from the controlling system can be applied to any
part of the controlled system.
The main application field for this kind of model is
research on body temperature regulation itself. No model
has been developed which also predicts whether a par-
ticular thermal environment is thermally uncomfortable
and to what degree. It may be possible to link a model of
this kind with the present knowledge on temperature
sensation and thermal comfort, so as to enable comfort
predictions to be made for transient conditions. This is
however beyond the scope of the present study.
From the above discussion it follows that at present
there is no other source except results of thermal comfort
experiments to assess the acceptability of changing
environmental conditions.
3. EXPERIMENTS
A large number of experiments have been conducted
on man's response to the thermal environment. Con-
cerning the objectives of the experiments, distinction can
be made between investigations on the thermoregulatory
system on the one hand and the establishment of thermal
comfortable or acceptable conditions on the other hand.
The latter type of experiments are primarily of interest
in the present study.
Although most work has been concentrated on steady-
state conditions, some experiments have examined tran-
sient conditions. In principle any of man's last heat bal-
ance variables (T. and T, or T0 , v, rh, Mand lc 1) may
change in time. However in most cases, changing ambient
temperatures has been of interest. Changes can be cat-
egorised as :
e Cyclical: triangular or sinusoidal changes in the tran-
sient variable (e.g. resulting from the deadband of the
HY AC control system), characterised by mean value,
peak to peak amplitude and fluctuation period or
frequency*
e Ramps or drifts : monotonic, steady changes with
time. Ramps refer to actively controlled changes and
drifts to passive changes (as one might encounter in
a building with no active temperature control). These
changes are characterised by starting value, amplitude
and rate of change ~
e Steps, such as one experiences in going from o'n.e
thermal environment to another. Step changes are
described by starting value, direction and amplitude.
J
The following section describes the results of the most
important thermal comfort experiments with cyclical
temperature changes since these are primarily of interest
in the present context. The next section describes results
of some other related experiments. All results relate to
environmental conditions in or near the comfort zone
for sedentary or slightly active persons wearing normal
indoor clothing.
·:
3.1. Cyclical temperature changes
Sprague and McNall (20] conducted experiments
aimed at providing data, obtained under controlled con-
ditions, as a basis for confirming or modifying existing
specifications on fluctuating thermal conditions. Before,
these specifications were largely based on field experience.
~1
312 J. L. M. Hensen
Their first series of tests were designed to study the effect
of fluctuating dry bulb temperature on the thermal
sensation of sedentary persons (N = 192; college age;
M= 1.2 met; / 01 =0.6 clo; T,=25.6°C; rh=45%;
v.< 0.15 m s- 1). The dry bulb temperature varied accord-
ing to a triangular wave form with average fluctuation
rates in the range l.7 to 10.9 K h- 1 and peak to peak
amplitudes ranging from 0.6 to 3.3 K, resulting in l.O to
2.0 cycles h- 1• All tests started from the middle of the
comfort zone (mean dry bulb temperature was 25.6°C) .
Although it is not clear how acceptability was defined
the authors concluded that no serious occupancy com-
plaints should occur due to dry bulb temperature fluc-
tuations if t.T;,P · CPH < 4.6 K 2 h- 1 in which ATp•p is
the peak to peak amplitude of the temperature fluc-
tuation and CPHis the cycle frequency (cycles h- 1) . This
expression, which was only validateq. inside the comfort
zone and for two fluctuation rates, suggests that ATP•P
could be large for s.low fluctuations and that ATP'P would
have to be small when fluctuations are rapid. This result
looks strange; when the human body is regarded as one
or more thermal capacitances, one woWd expect opposite
results (i.e. increasing acceptable ATP•P with increasing
fluctuation rate) . Therefore, results like this must be
related to the thermoregulation control mechanisms and
indicate that the rate of change of temperature is very
important.
The authors specifically state that their expression does
not apply to systems where the mean radiant temperature
fluctuates, since the effect of varying radiant temperatures
was not investigated. However, assuming [l 7) that, at
air speeds of0.4 m s- 1 or Jess the operative temperature
is simply the arithmetic mean of dry bulb temperature
and mean radiant temperature, t11e relation between
maximum acceptable peak to peak amplitude and cycle
rate of operative temperature can be assumed to be
!J.T;,P·CPH< l.2K 2 h- 1•
Following these results the tolerated range of tem-
peratures decreases with increasing fluctuation rates. This
seems to be contradicted by work of Wyon et al. (21]
who performed experiments in which the amplitude of
the temperature swings was under the subjects' control.
They found th.at subjects tolerated greater amplitudes
when the temperature changed more rapidly. In their
view this was due to purely physical reasons, as rapid
changes of ambient temperature cause skin temperature,
and hence lhermal sensation to lag further behind in time
and this effectively reduced the experienced temperature
fluctuations. It. was also found that subjects tolerated
greater amplitudes when performing mental work than
when resting. Mcintyre and Griffiths [22] later pointed
out that due to a much smaller rate of change of the
mean radiant temperature, when compared with the air
temperature, and unusual acceptability criteria (spon-
taneous dial voting when the temperature was too hot or
too cold) the tolerated range in operative temperature
was actually smaller than normally found in steady-state
conditions.
Later experiments of Wyon et al. [23) were designed
to investigate the effects on comfort and performance of
predetermined ambient temperature swings under more .
normal working conditions. The subjects (N = 16 ; stud-
ent age ; M = 1.2met ; 101 =0.6 clo; v < 0.1 m s - 1 ) were
exposed to sinusoidal swings around the average pre-
ferred ambient temperatures with peak to peak ampli-
tudes in the range 2-8 K and periods ranging from 32 to
8 min (i.e. 1.9-7.5 cycles h- 1) , resulting in fluctuation
rates between 15 and 60 K h- 1• Certain complications
resulted in considerable damping (up to 75%) of the
amplitude of the temperature swings below head level.
Also the actual amplitudes of the mean radiant tem-
peratures were lower than half of the intended ampli-
tudes. The authors state that for these reasons the experi-
ments are probably best regarded as an investigation of
air temperature swings at head height. From the results
they concluded: "Large temperature swings ... cause
increased discomfort" and "Large ambient temperature
swings appear to have a stimulating effect that is to
be preferred to the apparently opposite effect of small
temperature swings, but a constant, optimally comfortable
temperature, where this can be achieved, would still seem
to be preferable to either''. To be able to compare these
results with the other references, Wyon's raw data was
examined. This revealed that 80% of the votes were in
the comfort zone for all swings with intended peak to
peak amplitudes of 4 K or less. As indicated above this
actually suggests maximum acceptable peak to peak
amplitudes of operative temperature fluctuation for the
whole body in the range 1-2 K.
Experiments with large ambient temperature swings
were also conducted by Nevins et al. [24]. The subjects
(N = 18 · different ages; M = 1.2 met · lc1 = 0.6 clo;
rh = 50% ; v = 0.25 m s- 1) were exposed to ambient
temperature (T. = T,) swings with a peak-to-peak ampli-
tude of 10 Kand an average fluctuation rate of 19 K h- 1
(0.9 cycles h- 1). The mean ambient temperature was
25°C. From the results it was concluded that the preferred
ambient temperatures for comfort agreed well with the
results of earlier steady-state experiments (on which, for
instance, ASHRAE [25) is based) and that there was
no clear evidence of an increased or decreased range of
acceptable ambient temperatures due to fluctuation. An
examination of Nevins' raw data however suggests a
maximum acceptable peak-to-peak amplitude of about
2.8 K. This is a little less than the width of the comfort
zone for steady-state conditions. It should be noted that
when unacceptable temperatures are left out, a rate of
temperature change of 19 K h- 1 would have resulted
in a fluctuation frequency of about 3.4 cycles h- 1 or
alternatively 0.9 cycles h- 1 would have resulted in an
average rate of change of 5 K h - 1•
Robles et al. [26) conducted a series of experiments in
which the subjects (N = 804 ; college age· M = 1.2 met ;
J01 = 0.6 clo; rh = 50%) were exposed to cyclical changes
around various basal temperatures (17.8- 29.4°C) with
different amplitudes (l.I to 5.6 K) at rates ranging from
I.I to 4.4 K h- 1 (OJ to 1.5 cyc.les h- 1). The results
showed that if (steady-state) temperature conditions for
comfort are met, the thermal environment will be accept-
able, for near-sedentary acLivity while wearing summer
clothing, if the rate of changes does not exceed 3.3 K h- 1
and the peak-to-peak amplitude is equal to or less than
3.3 K (which is approximately the same as the width of
the steady-state comfort zone). The discussion following
the presentation of the results revealed some criticism
which was acknowledged by the authors. Apparently,
 Literature Review on Thermal Comfort
their acceptability criteria were less coarse than usual.
Due to the heat capacity of the building fabric, the mean
radiant temperature swings were damped and delayed
when the air temperature cycled. For this reason the
acceptable maximum rate of change and peak-to-peak
amplitude of operative temperature will probably be
lower than the values mentioned above.
There are a number of difficulties which should be
noted when comparing the results of the above men-
tioned experiments :
e The results are in fact subjective responses of a highly
complex system of which we most probably do not
yet know all the processes involved to the extent
necessary for controlling all relevant parameters dur-
ing experiments
e Usage of different semantic voting scales, both in type
(i.e. directed towards acceptance (with words like
acceptable and unacceptable), comfort, sensation or
mixed) and appearance (e.g. 2, 7 or 9 point, and
discrete or continuous)
e Differences in acceptability criteria (e.g. comfort
interval on a 7 point semantic comfort scale defined
as centre-point ± 1.0 vote as opposed to centre-point
±0.5 votes) which is sometimes unavoidable because
of the scale differences
e Differences in conditions: subjects resting or per-
forming mental work, fluctuating dry bulb tem-
perature or fluctuating operative temperature
e Differences in subjects; our knowledge of the dis-
tribution ofthermoregulatory efficiency (and thus the
time factor in discomfort) among people is still very
limited and this can easily lead to sample errors.
Regardless of these differences all results seem to indicate
that with cyclical fluctuating ambient temperatures the
bandwidth of acceptable temperatures decreases with
increasing fluctuation frequency. This bandwidth seems
to be at its maximum in steady-state conditions. This can
be seen in Fig. 2 which comprises the major results of the
experiments and indicates which fluctuation frequencies
were investigated.
The results suggest that there is a certain amplitude
threshold (at about l K) below which the influence of
fluctuation frequency is negligible. At frequencies below
approximately 1.5 cycles h- 1 the maximum acceptable
peak-to-peak amplitude increases with decreasing fre-
quency until the steady-state comfort bandwidth is
reached.
As shown in Fig. 2 the results seem to be quite
adequately described by ASHRAE's standard 55 [7]
which states with regard to cycling temperature: "If the
peak variation in operative temperature exceeds 1.1 K the
rate of temperature change shall not exceed2.2 K/h. There
are no restrictions on the rate of temperature change if
the peak-to-peak is l. l Kor less." The maximum rate of
temperature change of about 2.2 K h- 1 can be regarded
as conservative when compared with the experimental
results.
3.2. Other changes
Comfort experiments involving temperature drifts or
ramps are reported by Mcintyre and Griffiths [22], Berg-
lund and Gonzalez [27, 28], Berglund [29] and Rohles et
313
al. [30]. From the results it may be concluded that slow
temperature changes up to about 0.5 K h- 1 have no
influence on the width of the comfort zone as established
under steady-state conditions.
Mcintyre and Griffiths [22] report no difference
between temperature changes of 0.5, I .0 and 1.5 K h- 1
nor steady-state with respect to permissible deviations
from neutral temperature.
Berglund and Gonzalez [27] found however that with
faster rates of temperature change (i .e. 1.0 and 1.5 K
h - 1) the permissible deviation from neutral temperature
was larger than was the case for the 0.5 K h- 1 tem-
perature change. This difference was more pronounced
for subjects wearing summer clothing (0.5 clo) than for
those wearing warmer clothing (0.7 or 0.9 clo). It should
be mentioned however that these authors used an unusual
assessment of acceptability. Instead of the more common
procedure of deriving acceptability indirectly from
comfort votes, a direct two point acceptability question
was used. This resulted in a considerably wider ambient
temperature zone where the acceptability of the subjects
was 80% or higher when compared to the usual comfort
zones. Also the acceptable zone was shifted somewhat to
the warm side, implying that a slightly warm environment
is more acceptable than a slightly cool one.
From their eight-hour-long experiments Berglund and
Gonzalez [28] concluded that a temperature ramp of 0.6
K h- 1 between 23 and 27°C was thermally acceptable to
more than 80% of the subjects (wearing summer cloth-
ing). This would imply an increased comfort zone. The
section on temperature drifts or ramps in the ASHRAE
standard [7] states that" ... slow rates of operative tem-
perature change (approximately 0.6 K/h) during the occu-
pied period are acceptable provided the temperature during
a drift or ramp does not extend beyond the comfort zone
by more than 0.6 K and for longer than one hour." This
statement is most probably based on these results. As
indicated above, the results are however based on a
different acceptability assessment from the usual ones.
Furthermore, as Benzinger [13] points out, the results
may have been influenced by the fact that man's ther-
moregulatory set point is higher in the afternoon than in
the morning; that is, our tolerance for heat increases
during the day. In view of this, the ASHRAE standard
[7] should probably be restricted to acceptable changes
during daytime and in upward direction only.
From Nevins' [24] experiments with cyclical changes
with average fluctuation rates of 19 K h- 1 it was con-
cluded that fuere was no clear evidence of increased or
decreased comfort zones due to fluctuation of ambient
temperature. As pointed out by Mcintyre and Griffiths
[22], the results of the experiments with about the same
average fluctuation rate by Wyon et al. [21] on the other
hand do seem to provide evidence of decreased acceptable
ranges due to fluctuation.
From experiments in the 1950s by Hensel (also ·
reported in [l l]) it became clear that when the human
skin is exposed to changing temperatures the difference
between neutral temperature and the temperature at
which warm or cold sensations occur (i.e. thermal sen-
sation threshold) decreases inversely with the rate at
which the temperature is changed. This thermal sensation
threshold depends also on the temperature to which the
·1
I '

314 J. L. M. Hensen

'.•I 5.0
x Sprague et al. (experiments)
, '· Sprague et al. (expression)
o Wyon et al.
6 Nevins et al.
4.0 - o Rohles et al.
~~- ASHRAE Standard 55-1981

Peak to peak3 · O - ·
amplitude [K]
'
-. 2.0 -

0 a

.
' !(-
rr • 1.0 - ..
• , _-&---
-~~~~~~~~~~~~~~~~---1

. ...
'.
')(
. .. ....... .... . . .
............. .. . . . ........... . ... .... .

0.0
0.0 2.0 4. 0 6.0 8.0 10.0

Frequency [cycles/hour]
Fig. 2. Maidmurn acceptable peak-to-peak amplitudes of cyc.lical fluctu ating operati ve temperature as a
funct ion of cycle frequency for near-seden tary activity while wearing summer clothing (derived from
Sprague and McNall Jr (20),• Wyon (23], • Nevi ns et al. [24]. Rohl es et al. [26]. ASH RAE (7)t) .
• Operative temperatures estima ted from given dry bu.l b temperaLUres (see text).
t Value at 0.0 cycles 'b- 1 indicates width of steady-state comfort band.

skin is adapted when the change starts, on the direction
of change, on the exposed part of the body and on the
area being exposed. The latter two factors have a con-
siderable influence on the intensity of temperature sen·
sation as well. Although it cannot be proved, these
aspects may very well be partly the cause of the con-
tradictory results and conclusions of the experiments dis·
cussed above. The fact that there is a threshold for ther-
mal sensations, and that this threshold is affected by the
rate of temperature change, makes it likely that the same
is true for thermal comfort. This would be in support of
Fig. 2.
Contradictory results are also found with respect to sex
differences. Wyon et al. [31}, using high-school pupils,
found significant differences between the responses of
male and female subjects when exposed to changes in
ambient temperature (about 4 K h- 1 ) . Males in general
feel hotter and react faster than females. Nevins et al.
[24}, using college age males and young and older female
office workers, reported that the females had significantly
higher warmth sensitivity than the male group.
An explanation for these and previously mentioned
contradictions may be related to the choice of subjects
(i.e. sampling error). This can be deduced from the con-
clusion of Stolwijk [32} who, after reviewing a con-
siderable amount of research in this area, states : "Differ~
ences in effectiveness of the thermoregulatory system in
different individuals will result in different dynamic comfort
responses to changing thermal environments: people with
efficient thermoregulation will experience thermal dis·
comfort sooner than those with less effective thermo-
regulatory systems. Our knowledge of the distribution of
thermoregulatory efficiency among people is still very
limited."
The effect of the level of clothing insulation and activity
on man's thermal sensitivity during temperature changes
was investigated by Mcintyre and Gonzalez [33]. They
exposed young college males who were either heavily
clothed (1.1 clo) or almost nude and who were either
resting (1.1 met) or bicycling (2.3 met) to a 6 K step
change in air temperature. The temperatures were so
chosen that the subjects started warmer than neutral and
finished cooler than neutral. The experiments took place
in June and where partly replicated in August (after
summer heat acclimatization) to see whether there are
seasonal changes in thermal sensitivity. From the results
it was concluded that in general the change in whole body
thermal sensation was affected by clothing, exercise and
season. For resting subjects thermal sensitivity was not
affected by clothing insulation or season. However the
change in skin temperature following a change in air
temperature was greater when unclothed than clothed.
From this the authors concluded that change in mean
skin temperature is therefore not an adequate predictor
of thermal sensation. For unclothed subjects thermal
sensitivity was greater when resting than when exercising.
The responses of clothed, exercising subjects interacted
with season (e.g. they felt cooler in August).
As indicated earlier, the effect of greater sensitivity
during rest than when performing mental work was also

I·
 Literature Review on Thermal Comfort 315

found with the cyclical temperature change experiments
by Wyon et al. [21]. That clothing insulation does not
seem to have an effect on thermal sensitivity may be
explained by the fact that in general various thermally
sensitive parts of the body (e.g. hand, neck, hands) are
uncovered.
Probably because of the minor influence of moderate
humidities on thermal comfort and tbermal sensation,
there are only few experiments reported which investigate
the effect of changing humidity. Four studies, those by
Gonzalez and Gagge [34] , Nevins et al. [24], Gonzalez
and Berglund [35] and Stolwijk [32] all indicate that when
operative temperature is inside or near the comfort zone,
fluctuations in relative humi.dity from 20 Lo 60% do not
have an appreciable effect on the thermal comfort of
sedentary or slightly active, normally clothed persons.
Relative humidity becomes more important when con-
ditions become warmer and thermoregulation depends
more on evaporative heat loss.
Regarding changing air velocities no references have
been found except of course those dealing with the effect
of air turbulence on sensation of draught. Velocity fluc-
tuations due to turbulence are in general much faster
(ranging from 0.01 to 10 Hz) than ambient temperature
fluctuations which generally can be measured in units of
cycles h- 1• Fanger et al. (36] concluded that an air flow
with high turbulence causes more complaints of draught
than air flow with low turbulence at the same mean
velocity. As possible reasons for this were mentioned the
relation between convective heat transfer and turbulence
and the relation between the heat flux (or rate of tem-
perature change) as sensed by the skin thermoreceptors
and turbulence.
Finally it is repeated that care must be taken in apply-
ing the above results. [n general many contradictory
results have been found. These were most pronounced
with respect to rate of temperature change, sex difference
and age difference. The possible reasons are already indi-
cated in the previous sub-section.
4. CONCLUSIONS
Our theoretical knowledge concerning tbermal com•
fort in transient conditions is still limited. At present,
results of thermal comfort experiments seem to be the
only source of information on thermal acceptability in
changing environmental conditions.
The present study is restricted to conditions charac-
teristic for homes, offices, etc. The (ollowing conclusions
are supplementary to the steady-state comfort criteria
which are usually associated with those conditions; i.e.
sedentary or slightly active persons, wearing normal
indoor clothing in an environment with low air move-
ment ( < 0.15 m s- 1) at 50% relative humidity.
The experimental results related to cyclical fluctuating
ambient temperatures are, although perhaps a little con-
servative, quite adequately described by ASHRAE's
standard (7] which states with regard to cyclic changes:
"If the peak variation in operative temperature exceeds
1.1 K the rate of temperature change shall not exceed 2.2
K/h. There are no restrictions on the rate of temperature
change if the peak-to-peak is I. I K or less."
With respect to tempera ture drifts or ramps, there is
good experimental evidence that at rates of operative
temperature change below 0.5 K h- 1, the environment is
experienced as in steady-state condhions. At rates
between 0.5 K h- 1 and 1.5 K h- 1 there is, apart from
experiments with uncommon acceptability assessment
procedures, no clear evidence of increased or decreased
comfort zones due to transient conditions. The para-
graph in ASHRAE's standard [7] states that " ... slow
rates of operative temperature change (approximately 0.6
K/h) dul'i11g the occupied period are acceptable provided
the temperature during a drift or ramp does 1101 extend
beyond the comfort zone by more than 0.6 Kand for longer
titan one hour", but this should probably be restricted
to acceptable changes during daytime and in an upward
direction only. No evidence was found why the limit for
cyclical changes (i.e. if the rate of temperature change
exceeds 2.2 K h - 1 the peak variation shall not exceed 1.1
K) would not be valid for temperature drifts and ramps
as well as for higher rates of change of operative tem-
perature.
From several experiments it was found that clothing
insulation has a negligible effect on thermal sensitivity
during temperature changes. This implies that the limits
stated above are valid for summer as well as winter con-
ditions.
Regarding activity level, a greater sensitivity was gen-
erally found during rest than when performing mental
work. From this it follows that the above Limits may be
regarded as conservative in case oflighl sedentary activity
in offices, homes, e~c. ' .
Provided operative temperature is inside the comfort
zone, humidity fluctuations, when relative humidity is in
the range from 20 to 70%, do not seem to have an
appreciable effect. Regarding changing air velocity, no
references were fourld except those dealing with the effect
of increased draught complaints when air turbulence is
higher.

REFERENCES
I. ISO. Moderate thermal environments--<letennination of the PMV and PPD indices and specification
of the conditions for thermal comfort. International Standard ISO 7730, International Organisation
for Standa.rdization (1984) .
2. P. 0 . Fanger, Thermal Comfort. Analysis and Applications in Environmental Engineering. McGraw-
Hill, New York {1972).
3. D. A. Mcintyre, Indoor Climate. Applied Science, London ( 1980).
4. A. P. Gagge, A. P. Fobelets and L. G . Berglund, A standard predictive index of human response to
the thermal environment. ASHRAE Trans. 92: 28. 709-73 l (1986).
5. A. P. Gaggc, Rational temperature indices of man's thermal environment and their use with a 2-node
model of his temperature regulation. Fedn Proc. Fed. Am. Soc. Exp. Biol. 32, t572- 1582 (t973).
6. P. 0 . Fanger, Proposed Nordic standard for ventilation and thermal comfort. Proceedings of the
International Conference on .Building Energy Management, Povoa De Varzim, Portugal (1980).
 ~ --- .. ~1-----·~~..........................~........lilllliilllil. . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . .. . . ...

316 J. L. M. Hensen

. ,,.. 1 :: it7~n ASHRAE, Thenn.111 environmental condi tions for huma n occupancy. ANSI/ASHRAE Standard 55-
: .. Yf!''I ' 1981, American Society of Heating, Refrigerating and Air-Conditioning Engineers,, Atlanta (1981).
8. -· M. V. Jokl, A new "COMECON" sta·ndard for thermal comfort withi n residential and civic buildings.
;:,, "".'" Indoor Air '87, Vol. 3, pp. 457-460, Berlin (1987).
- 9!L;T, L. Madsen, Measuremen t and control of thermal comfort in passive solar systems. Proceedings of
: ' ,, _,;;, !It \1" the 3•d In ternational Congress on Bui1ding Energy Management ICBEM '87 (edited by A. Faist, E, 1,,,. ••• ! "'
.. .. ., ; : Iii ~· <' 'J.Femandes and R. Sagelsdorlf), Vol. IV, pp. 489-496, Presses Polytechniques Romandes, Lausanne
" :; ; ' (1987).
· · 10. · J. L. M. Hensen, L. C. H. Dings and W. J. A. van der Ven, Meetproject Someren. Technische
· ; ' Universitei t Eindhoven rapport FAGO 87.08.K., Eindhoven (1987).
1V H. Hensel, Thermoreception and Temperature Regulation (Mongr . physiol. Soc . 38). Academic Press, · ·
~ · ~.- • .,,, , . H: :,.. London (1981). : I "I
12. T. L. Madsen, Why low air velocities may cause thermal discomfort? Proceedings of the 3n1 Inter-
national Conference on Indoor Air Quality and Climate. Stockholm, pp. 331- 336 (1984) .
13. T. H . Benzinger, The physiological basis for thermal comfort. Indoor Climate (Edited by P. 0 . Fanger
.,. and 0. Valbjorn), pp. 441-476, Danish Building Research Institute, Copenhagen (1979).
14. A. P. Gagge, J. A. J. Stolwijk and J. D. Hardy, Comfort and thermal sensa.tions and associated • : I ~ a

,! responses at various ambient temperatures. Envir. Res. 1, 1- 20 (1967).

15. ,. J. D. Hardy, Thermal comfort : skin temperature and physiological thermoregulation. Physiological
· a11d Beliaviara/ Temperature f._egulat(on (Edited by J. D. Hardy, A. P . Gagge and J. A. J. Stolwijk),
pp. 856-873, Thomas.Books, Springfield, IL (1970).
16. · ASHRAE, Handbook of F1mdame1uals. American Society of Heating, Refrigerating and Air-Con-
i- ditioning Engineers. Atlanta (1985).
J7. T. J. Doherty and E. Arens, Evaluation of the physiological bases of thermal comfort models. ; ..
: " ." ASHRAETra11s. 94:1, in press. r.
18_- J. A. J. Stolwijk, Mathematical model ofthermoregulation. Physiological mid Behavioral Temperature
' ' ' .- RegulatioM (Edited by J. D. Hardy, A. P. Gagge and J. A. J. Stolwijk), pp. 703- 721 , Thoma.s Books,
Springfield, IL ( 1970).
19. R . G. Gordon, The response of a human temperature regulatory system model in the cold. PhD
Thesis, Santa Barbara, University of California (1974).
20. C.H . Sprague and P. E. McNall Jr, The effects of Ouctuating temperature and relative humidity on .
the thermal sensatfon (thermal comfort) of sedentary subjects. ASHRAE Trans. 76 :1, 146-156 (1970) .
21. D. P. Wyon, N. 0 . Bruun, S. Olesen, P . Kjerulf-Jensen , and P. 0. Fanger, Factors affecting the
subjective tolerance of ambient temperature swings. Proceedings of the 5•h International Congress for
Heating, Ventilating and Air Conditioning, Copenhagen (Edited by H . Salmark), Vol. 1, pp. 87- 107
(1971 ).
22. D. A. Mcintyre and I. D . Griffiths, Changing temperatures and comfort. Bldg Serv. Engr 42(8), 120-
122 (1974).
23. D. P. Wyon, T . Asgeirsdottir, P. Kjerulf Jensen and P. O. Fangec, The effects of ambient temperature
swings on comfort, performance and benaviour. Arch. Sci. Physial. 27A, 441-458 (1973).
24. R. G. Nevins, R.R. Gonzalez, Y. Nishi and A. P. Gagge, Effect of changes in ambient temperature
and level of humidity on comfort and thermal sensations. ASH RAE Trans. 81 :2, 169- 182 (1975).
25. ASHRAE, Thermal comfort conditions for human occupancy. ASHRAE Standard 55-1974, Ame.r-
ican Society of Heating, Refrigerating and Air-Conditioning Engineers, New York (1974).
26. F. H . Rohles , G . A. Milliken, D. E. Skipton and I. Krstic Thermal comfort duringcyolical temperature
fluctuations . ASHRAE Trans. 86:2, 125-140 (1980).
27. L. G. Berglund and R. R. Gonzalez, Application of acceptable temperature drifts to build environ-
ments as a mode of energy conservation. ASHRAE Trans. 84 :1, 110-121 (1978).
28. L. G. Berglund and R. R. Gonzalez, Occupant acceptability of eight-hour-long tempera ture ramps
in the summer at low and high humidities. ASHRAE Trans. 84:2, 278-284 (1978).
29. L. G. Berglund, Occupant acceptance of temperature drifls. Indoor Climate (Edited by P. 0. Fanger
and 0 . Valbjorn), pp. 507- 525, Danish Building Research Institute, Copenhagen (1979) .
30. F. H. Rohles, J. E. Laviana, R. Wei and R. Wruck, The human response Lo temperature drifts in a
simulated office environment. ASHRAE Trans. 91 :IA, 116-123 (1985).
31. D. P. Wyon, I. Andersen and G. R. Lundqvist, Spontaneous magnitude estimation of thermal
discomfort during changes in the ambient temperature. J. Hygiene 70, 203-221 (1972).
32. J. A. J. Stolwijk, Physiological responses and thermal comfort in changing environmental temperature
and humidity. Indoor Climate (Edited by P. 0. Fanger and 0. Valbjorn), pp. 491-506, Danish
Building Research Institute, Copenhagen (1979).
33 . D. A. Mcintyre and R. R. Gonzalez, Man's thermal sensitivity during temperature changes at two
levels of clothing insulation and activity. ASHRAE TraMs. 82 :2, 219-233 (1976) .
34. R. R. Gonzalez and A. P. Gagge, Magnitude estimates of thermal discomfort during transients of
humidity and operative temperature and their relation to the new ASHRAE Effective Temperature
(EP) . ASH RAE Tra11.r. 79:1 , 88- 96 (1973) . .
35. R. R. Gonzalez and L. G. Berglund, Efficacy of temperature and humidity ramps in energy conser-
vation . ASH RAE J. 6, 34-41 (1979) .
36. P. 0 . Fanger. A. K. Melikov, H. Hanzawa and J. Ring, Air turbulence and sensation of draught.
Energy Bldgs 12, 21 - 39 (1988) .